Class A, AB or what?....

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Ok, here's my dilema. I'm about to build a stereo integrated preamp/amplifier to start forming my hi-fi rig. I have the preamp already designed, and i was looking for power amplifiers plans on the web to work on. Right now i'm between any of the "Zen" derivates by Nelson Pass and the "New improved 60-100W amp" by Rod Elliot ( Here's my dillema: everyone claims the Zen series sound incredible, and better than comparable ab amps. How came an amp that has higher THD and worst frequency response/bandwith sound any better (even if the distortion is almost completely 2nd harmonic)? Besides, if i finally go for a class-A, i won't be needing much more than 40-60W per channel, so which design would you recommend that would sound good and dissipates the least amount of heat? Tnx.
I too am in a similar sort of dilemma.But I think that you could try the 60W amp from ESP.It claims to have a good THD figure and has reasonable power.I do not say that the Zen may sound bad due to its THD figure.The main reason these Class A push-pull or single ended sound good is due to the fact that they are biased in Class A.Valve amplifiers do not always have excellent THD figures but they are supposed to sound good.The great Williamson amplifier is supposed to have a bad THD figure,but it is considered as one of the best ever designed.As far as heat dissipation is concerned,Class A push-pull will dissipate less heat than Class A single ended.Besides,the sound from your audio system is not solely dependent on the performance of your amplifier.
You could have a look at the Hood Amplifier at
Good luck and keep me informed of any developments.
I think i'll go with Elliot's design; for what i've been checking it seems to be a very good amplifier.

On the other hand, i've been thinking about the class-a issue lately, and i beleive the reason class-a amps are "good sounding" is because they add "pleasurable" distortion (2nd harmonic); i still dunno if that could be considered Hi-Fi, but's the only explanation i can found of why an amp with much higher THD could sound better.
Class A amps do not add 2nd harmonic distortion. They are inherently more linear because there is no crossover distortion. The output transistors are always on unlike in a class AB amp, where each half of the push/pull pair switches on and off on each cycle.

The zen amp has higher THD figure because it has only a single gain stage and no negative feedback.

There are a couple of good class A designs on the web that use negative feedback and have very low THD figures: Rod Eliott has a 20w class A based on his 60w one, and Pass has three papers on class A amps, one at 20w, one at 40w, and one at 75w. Be prepared to spend big bucks on heatsinks if you do more than 20w or so.

One thing you might try: modify the designs so that they all run off the same supply voltages and use approximately the same size heatsinks. Then you can build both amps and just drop the circuit boards into your chassis to try out. Transistors are cheap and are usually less that 10% of the total amplifier price when you consider how much transformers, heatsinks, chassis, etc. cost.

Now, this is my problem.... what's the good of being more "inherently linear" if the THD is higher. I mean, i picture the amp as a black box; any non-linearity is harmonic distortion added to the amplified signal right? So, if the class ab amp has higher bandwidth, better frequency response, and lower THD, how can it sound worst?

I once readed class-a designs are more prone to add 2nd harmonic distortion; that *could* be an explanation, but i sincerely dunno.
By inherently more linear, I mean more linear before the application of global negative feedback. THD is a rough measurement of linearity in the final amplifier, and by that measure, the Rod Eliott design is much more linear than the zen.

The zen only has one amplification stage and no negative feedback. In theory, you could lower any amp's distortion by using some of the amp's gain for feedback, but the design as it stands has very little gain (8.5dB) so there isn't much flexibility here.

Taking a wild guess, the Rod Eliott design has an open loop gain of around 100dB or so, more if you use a constant current source in the gain stage, and you probably want the final closed loop gain for the amp to be around 30db, leaving around 70db for feedback. This reduces the THD and output impedance by a factor of around 3000. So if you used the Eliott 60w amp without feedback, it would probably have a THD figure somewhere around 20-30%!

This is what I mean by "intrinsically more linear." The Zen, without feedback, remains quite listenable (I haven't heard it myself, but most reports are glowing), while the Eliott amp would sound terrible. However, because the Eliott amp has so much more gain, it can use it, via feedback, to end up much more linear in the end.

So what's the advantage of a single-stage amp like the zen, besides using few parts? Why don't we just build amps with 1000 dB open loop gain and use massive amounts of feedback to bring the THD number down to unmeasurable levels?

Well, that's exactly how they built them in the '70s and they sounded like crap. Why? Because THD isn't everything. The more feedback you use, the more susceptible your amp becomes to instability (which may cause it to oscillate or ring during transients), and Transient Intermodulation Distortion (TIM), which is the inability to reproduce fast-moving waveforms accurately. Both are harder to measure than THD.

The current trends in amp design are all tending towards using inherently linear circuits and then using moderate amounts of negative feedback (20dB or so) to raise the damping factor and lower the THD even further. The "Leach Amp" is a good example of this sort of design.

The Zen and the Eliott amp embody extremes of this debate. The Zen is extremely "new school:" high intrinsic linearity, low open loop gain, little (in its case, no) feedback. The 60w amp is decided "old school:" high open loop gain, lots of feedback, and likely pretty awful intrinsic linearity. The Leach amp is somewhere in the middle, as are Pass's other class-A designs.

The push for Class A is not that it produces more 2nd harmonics than AB, but that it pretty much produces ONLY 2nd harmonics, especially with highly linear devices like MOSFETs or Tubes. Class AB amps will produce more higher order distortions because of switching noise at the crossover point.

If you use fully balanced, push-pull circuitry in class A, the two halves of the circuit will cancel much of the 2nd harmonic distortion, leaving you with a low THD number even without feedback. I believe this is one of the principles underlying Pass's new "X" series amps, but his patent is very opaque reading.

A nice looking design I've come across (and will hopefully build someday if I can find suitable heatsinks) is Jean Hiraga's "Le Classe A" (don't have my links, but Google will find it). It is a very simple (6 transisters, vs. zen's three) fully balanced 20w class A amplifier. An old design, but by a very highly regarded French designer.

The final thing to remember is that distortion is not the only indicator of sound quality. 2% second harmonic is not going to sound bad. In fact, studies have shown that a little bit of 2nd harmonic cause people to perceive the sound as more "pleasant" but less "accurate."

I would guess that if you like the sound of tube amps, you'd probably like the zen, since it's performance numbers are similar. On the other hand, if your ideal amplifier is a 300w Krell, then the Eliott design will get you a lot closer to what you want.

Hope this clears some things up.

Finnaly, a *GOOOD* answer! :) Tnx man.

You're right, and i never really thought on how much negative feedback affected the performance of an amp. Not only for THD (which seems to have become a huge issue lately), as i beleive THD is a "rare" phenomena, but more for the slew rate and stability.

Having said that, i still beleive h.distortion, even if 2nd harmonic, is *NOT* a good thing; i know it sounds "sweet", but if i wanted that, i would add it artifically myself (exciters do this, don't they?). Class A amplifiers with little applied feedback, can lower THD to very low levels without TIM and so, and thats basically what i'd like to build... without the heat :) The leach amp is nice, and i'm sure it sounds incredible, but way to complex, even to fit in an enclosure already containing switching, indicators, led vumeters and the preamps.

So, it's either a good ab amp (like the elliot, which in addition of very nice figures has about 80db gain open loop... would that be a good compromise?) or a low power class-a. What do you beleive?

BTW, the hiraga amp can be found at Hope that helps, and thanks again!
Can't really say. I'm sort of in the same position you are. I'm just getting back into the DIY electronics hobby after a long hiatus since high school. So, I'm also trying to decide which way to go.

Class A is pretty much orthogonal to low feedback in an amp design. There are very nice class AB designs that use low amounts (< 30dB, the leach amp is one of these) of feedback. There are also class A designs that use massive amounts (Krell).

I personally am fairly uninterested in the Eliott amp. It's a nice design, but it's very conventional and somewhat outdated. Part of the fun of DIY is that you can build something totally different than what you can buy.

Right now, I'm pretty enchanted by the hybrid tube/mosfet amp at this site: I don't think I'll build exactly this design, but something similar. The only problem with hybrids is that they require 3 separate power supplies!

He's also got the schematic for the "MJ Amp", which resembles the Hiraga amp, but with MOSFETs instead of transistors for the output.

One thing you might consider doing is building a roomy chassis, a solid power supply for whatever rated power you're interested in, and enough heatsinking. You could then substitute different amps into it and play around with a whole bunch of designs. Keep in mind that the price of the actual amplifier circuitry is typically no more than 15% of the total project cost, which is dominated by expensive transformers, chassis, and heatsinks.

As for my personal audio philosophy, it's something along the lines of:

1. moderate power (25-40w). This will drive my 90dB sensitivity speakers louder than I'll ever need.
2. THD less than 0.1%, but not obsessively low
3. minimal feedback (< 20dB)
4. simple circuits -- passive preamps, few gain stages, and the less active devices, the better.

Class A vs. AB doesn't really figure into it too much. Class A does sound better, but it's harder to get right and a pain to develop, and a well executed AB design comes close.

Let me know what you decide on!

First of all, where i said "... affected the performance of an amp. Not only for THD... as i beleive THD...", i was refering to transient intermodulation, not THD. Sorry.

Now, as you said, the fun of DIY is creating your own designs; unfortunately i'm still studying electronics so i'm limited to finding a good design and modifiying it, rather than starting from scratch... i was planning to do what you said to: building a suitable psu and getting a large heatsink and to try as many designs as possible. I just might if get the time for it, dunno...

I'm also looking for the same points you do on my audio amp. I don't think i'll be ever needing more than 40w per channel; low THD is a must; low feedback is also indispensable for a fast & free of TIM amp and well, i like my analog circuits as simple as possible too. To be honest, if i could find a low feedback class-a amp which wouldn't dissipate %300-400 the power it delivers, i'd work on it happily; but they don't exist. Tubes? ...I like their sound and have some very appealing characteristics, but tubes are a pain in the *** to deal with (fragile, limited life, microphonic, etc), so, i don't think i'll use them (but that's me).

I think i'll start working with the ESP amp and see where can i get upgrading it. It's an "old" design, i partially agree, but of all ab amps i've been checking it's the simplest and more documented design i've found, besides seeming to perform nice.

And, is anything wrong with Krell amps? :) I thought they were suposed to be great!

Once again, thanks.
Tubes are fun, less fragile than you might think. The high voltages and the need for separate heater supplies is a pain.

Have fun with ESP amp. It does look like a nice design and should be easy to get working well.

There's nothing wrong with Krell amps, they're just not what I go for. They are incredibly clear, powerful, and accurate...but they just don't involve me with the music.

My tastes run more towards the sound of British hi-fi gear -- small, polite, not the ultimate in extension, but musically very alive. Tubes are nice, too, but like you said, kind of a pain and major $$$.

What annoys me the most about solid state (SS) class A amps (Pass, Lindsey-Hood, etc) is the amount of heat generated an the incredible power consumption. ~90 watts idle power consumption for 15 watts amplifier are too much for me (see L-H design) plus they require a VERY sophisticated low noise power supply design. Just the weight of Al heatsinks and big mo-fo capacitors, transformers, inductors, and resistors make them an engineering challenge more than anything. IMHO these design is completely unsuited for integrated amp projects.
So far I have refused to build any SS class A, maybe one day I will build one just to see what the fuss is all about.

Here are just a few alternatives to SS Class A that come to mind:
-AB feedback, the US classic: HK citation; the EURO classics: Lecson (Bob Stuart of Meridian designed it), Radford. I built them all and work-sound great for integrated design.
-AB low feedback: Leach, Aloia volks.
-AB zero feedback, my favorite: Aloia GY-50, LCAudio End Millennium. I built an old Aloia zero feedback design from scratch back in 1988 used twentyfour (24) TO3 trans, it took me 2 years to build it but it's still working in my brother's system.

Good Luck!

[Edited by grataku on 01-02-2001 at 02:45 AM]
Thanks for the feedback. Right now i'm working on the ESP amp with a custom preamp and Rod Elliot's RIAA preamp (for phono input), complete with digital vu-meters, tune controls, selectable external preamp, headphones output, etc. I don't have as much spare time as i used to, so it won't be done soon. I'll let you know how it came when i finish it.

Trying to choose an amp is a hard task I know about that.
A class A amp sounds better because of it´s lower distortions in a lot of sections according to Douglas Self you can read about that in the book "Audio Power amplifier design handbook" or I think in his web site too.
A big BUT thow is what kind of speakers will you use with the amp. If they have a low sensitivity I don´t think you will get much out of the amp.
I have an alternative amp to propose.
The new hi end design from elektor magazine.
The improved edition uses bigger hexfets or IGBT´s in the output.
It´s a terific design and sounds even better.
My brother was shocked after listening to it because he uses a valve amp with ultralinear output.
It delivers 90 watts rms at 8 ohm.
The lab specs are excelent.
If anybody is interested I can send them the project in pdf. I made a small tweak in the project so I used a big power hexfet in the output instead of IGBT´s because i dont really trust them for sound quality.
Good luck with your project
Amen to that! :)

Yeah, i know what you mean. I basically choose the ESP amp because, besides having some terrific specs, it's simple and VERY well documented. I've no doubts about it sounding great, but it's a homemade proyect, and i have to balance the score somewhere :)

Btw, the speakers i'm using are a pair of Kenwood LS-70B i borrowed from my girlfriend (she was tossing them to the garbage!). I replaced both tweeters on them (one had the dust cover destroyed) and sound great on my dad's hi-fi. They're good, old speakers (tweeter+boomer, 60w max, 90db/wm sensitivity, large bandwidth, etc).

I'll check that url, tnx!
I recommend a class AB amplifier with a adjustable Vbe.
There are designs of amplifiers with an adjustable resistor to "set" more to class A than AB, or otherwise more to class AB than A. This can also be used when you're playing on a heat summer night: the transistors can run hot quickly; in this case you can set the amplifier more to class AB.

Did you consider the AKSA ?
Seems to me pretty much a good compromise between both worlds.
Or try - to be a little heretic - the NAIM AUDIO basic amplifier circuit. I know that it is somehow not PD, but the circuit is so basically simple and easy to construct. My NAIM clones work everyday perfectly.

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.